For centuries women have used abortifacient medicines,1 as evidenced by the Hippocratic oath proscribing their use: “… I will not give to a woman an abortive remedy.”2 Most of these botanical or chemical agents have not been researched systematically, but antimetabolite agents have been studied as abortifacients. In 1950, Thiersch and Philips3 reported fetal death in pregnant rodents injected with the folic acid antagonist 4-aminopteroylglutamic acid. Thiersch4 subsequently observed fetal expulsion in ten of 12 women less than 3 months pregnant who were given oral 4-aminopteroylglutamic acid. In 1981, Tanaka et al5 induced fetal death by injecting the folic acid antagonist methotrexate into an interstitial ectopic pregnancy. Stoval and Ling6 subsequently used intramuscular (IM) methotrexate for treatment of ectopic pregnancy.
Methotrexate also interrupts early intrauterine pregnancies. In women up to 42 days' gestation, Creinen7 noted that all aborted within 30 days after 50 mg/m2 IM MTX. Addition of the vaginal prostaglandin (PG) misoprostol after methotrexate induced more rapid abortion in some women.8 The antiprogestin mifepristone, when used with PGs, results in more than 95% rates of successful abortion in gestations up to 8 weeks after the last menstrual period (LMP).9 Unlike methotrexate, mifepristone has no proved effect in treating ectopic pregnancy and is less effective as a single-agent abortifacient, with success rates of 60–70%.10
An ideal abortifacient would be effective, inexpensive, orally administered, easy to store and transport, minimally toxic to pregnant women, and without teratogenic effects on future pregnancies. Mercaptopurine is a purine antagonist antimetabolite known to increase the risk of first-trimester abortion.11 Used to treat trophoblastic neoplasia,12,13 6-mercaptopurine might interrupt early ectopic and intrauterine gestations. 6-Mercaptopurine is inexpensive, orally administered, stable at room temperature, and minimally toxic.11 We compared oral 6-mercaptopurine and IM methotrexate, without PGs, for early medical abortion.
The mechanism of action of methotrexate in medical abortion remains unknown. Immunohistochemical cell-proliferation markers may provide insight into the process. The nuclear antigen Ki-67, expressed only in dividing cells, shows differing immunoreactivity in molar and nonmolar placentas, reflecting higher levels of cell division in the former.14 If abortifacients act by decreasing trophoblast proliferation, measured Ki-67 activity may be helpful in comparing potential abortifacients. Our analysis included Ki-67 activity in fetal tissues exposed to methotrexate or 6-mercaptopurine.
The University of Washington Institutional Review Board approved the study protocol. A free-standing abortion clinic served as the study site. Enrollment was initially offered to women who requested abortion with gestations less than 7 weeks from the LMP. We excluded women younger than 18 years, non-English speakers, those living more than 2 hours from the clinic, and those without a home telephone.
After consent for screening, a low-sensitivity (2000 mIU/mL) β-hCG urine test (UGC-Slide Test; Wampole Laboratories, Cranbury, NJ) confirmed pregnancy. If the result was negative, a high-sensitivity (25 mIU/mL) test was done (Clearview hCG II; Wampole Laboratories). Vaginal ultrasonography (Ultramark 4PLUS; Advanced Technology Laboratories, Bothell, WA) determined sac size or crown-rump length. Blood was drawn for quantitative β-hCG, complete blood count with platelets, creatinine, aspartate serum transaminase, and Rh factor. Exclusion criteria were white blood cell (WBC) count less than 3000/mL, hematocrit less than 30%, platelets less than 150,000/mL, creatinine greater than 1.5 mg/dL, or aspartate serum transaminase greater than twice the normal range (17–54 spectrophotometric units/L). In women without a visualized intrauterine gestation, enrollment was offered only for serum β-hCG less than 2000 mIU/mL.
Final exclusion criteria eliminated women with histories of ulcers, inflammatory bowel, or liver disease; those unable to comply with multiple clinic visits and blood sampling; those unable to abstain from alcohol and intercourse for 2 weeks; those taking folate-containing vitamins; those who refused suction abortion if medical abortion failed; those who were breast-feeding; and women who were seropositive for human immunodeficiency virus. The consent form emphasized the increased risk of congenital anomalies and the need for suction termination should the drug fail. Women agreed to suction termination if fetal cardiac activity remained after 14 days, with the option of electing suction termination at any other time.
For randomization, 50 note cards were designated methotrexate and 50 6-mercaptopurine. They were placed into opaque envelopes that were labeled with a unique 5-digit study number from a random-number table and then sealed. Each woman drew an envelope, and the investigator logged the study number and then opened the envelope. Allocation assignment was verified by the absence of missing or unopened envelopes at the conclusion of the study. Investigators and women were not masked to the drug received.
Methotrexate (Lederle Laboratories, Pearl River, NY) was given as 50 mg/m2 of body surface area, determined from a standard nomogram.15 The upper-outer quadrant of the gluteal muscle was chosen for methotrexate injection. Mercaptopurine (Burroughs Welcome, Research Triangle Park, NC) was given as four 50-mg pills swallowed in the clinic. If indicated, Rho (D) immune globulin (HypRho-D Mini-dose; Miles Inc., Elkhart, IN) was given. The women received prepackaged acetaminophen with 30 mg codeine (25 pills) and 600 mg ibuprofen (25 pills).
Each woman received a diary to record days with symptoms of abdominal cramps, dizziness, nausea, diarrhea, breast tenderness, headache, mouth sores, and fatigue. The diary included daily intake of acetaminophen tablets with codeine or ibuprofen. The women were given a 24-hour emergency contact number.
The women returned after 2 weeks for review of symptom diaries, nonsensitive urine pregnancy test, repeat blood draw, and vaginal ultrasonography. If fetal cardiac motion was present, the women underwent suction abortion. If no cardiac motion was present, an appointment was made for 2 weeks later. The women returned at 2-week intervals until vaginal bleeding occurred, ultrasound documented an empty uterus, and laboratory results documented a negative nonsensitive urine β-hCG test. In cases of missed appointments, the women were interviewed by telephone.
After enrollment of the first 27 randomized women, the protocol was modified with approval of the Institutional Review Board. 6-Mercaptopurine was no longer offered because of a 100% failure rate. Women were offered methotrexate only at a gestational age of less than 6 weeks because of more failures at later gestational ages. Screening laboratory tests were reduced to quantitative β-hCG, hematocrit, and Rh factor at enrollment only.
All suctioned tissue was submitted for pathologic evaluation. Representative sections of deparaffinized, formalin-fixed tissues were incubated with antibody MIB1 to the Ki-67 antigen,16 with heat epitope retrieval as described previously.17 Localization was by the avidin biotin immunoperoxidase technique.18 For quantification of Ki-67 indices, we used a microcomputer-based system for Ki-67–defined cell proliferation, as described previously.19 Indices from three random fields were averaged and the numbers were expressed in arbitrary units.
Student t test was used to compare means of continuous variables. Chi-square and Fisher exact tests were used when appropriate to assess categoric variables. All tests were two-tailed and all P values are reported.
From November 1995 to October 1996, 73 women were enrolled. The first 27 women were randomized to receive either methotrexate (n = 15) or 6-mercaptopurine (n = 12). Table 1 summarizes gestational age, level of β-hCG at entry, and demographic characteristics in the randomized groups.
No woman was excluded from enrollment by laboratory criteria. There were no differences in screening laboratory values between the randomized groups at enrollment. Within each treatment group, there were no changes in WBC counts, hematocrits, or platelets between visits 1 and 2. In the methotrexate group, aspartate serum transaminase increased from 18 to 23 spectrophotometric units/L (P = .03) by visit 2, still within the normal range. Mean β-hCG increased in the 6-mercaptopurine group from 21,418 to 101,792 mIU/mL (P < .001). In the randomized methotrexate group who successfully aborted, mean β-hCG decreased from 21,382 to 13,851 mIU/mL (P = .04).
Diaries were completed by 18 (66%) of the randomized women in each group. Table 2 summarizes the diary data. There were no differences in mean days of abdominal pain, cramps, or medications used between the groups, despite successful nonsurgical abortions in eight of nine women in the methotrexate group. The 6-mercaptopurine group differed from the methotrexate group by having a higher mean for days of fatigue.
Two women in the 6-mercaptopurine group returned on days 7 and 10 because of intolerable nausea and breast tenderness. Ultrasound showed fetal cardiac motion in both women, and suction abortions were performed. The remaining ten women in the 6-mercaptopurine group returned at or after the scheduled second visit (mean day of return 14.9; standard deviation [SD] 1.4). Ultrasound revealed positive fetal cardiac motion in all women, and suction abortions were done.
Immunohistochemical analysis of Ki-67 activity showed differing levels of cell proliferation in tissues exposed to 6-mercaptopurine and methotrexate. We analyzed tissue from four women in the randomized group exposed to methotrexate (two withdrawals, one incomplete abortion, one methotrexate failure) and 12 women exposed to 6-mercaptopurine (12 failures). Mean Ki-67 activity in methotrexate-exposed fetal tissues was 103.6 (SD 27.3), and in the 6-mercaptopurine-exposed it was 166.4 (SD 27.2) (P = .003). By the pathologist's masked qualitative assessment, there was no difference in the degree of hematoxylin and eosin staining between tissues with a marked difference in Ki-67 activity (Figure 1).
Sixty-one women received methotrexate alone, 15 randomized to methotrexate and 46 who received methotrexate after randomization was discontinued. Abortion without surgery occurred in 48 of 61. Abortion without surgery occurred in 45 of the 55 women who were less than 6 weeks after the LMP at the enrollment and in three of the six enrolled between 6 and 7 weeks after the LMP (P = .10).
Four of the 61 women (6.5%) withdrew from the study and requested surgical abortion. Reasons included not wanting to wait (suction on day 7 and day 14), relocation to a different city (suction on day 8), and a flulike illness believed to be exacerbated by pregnancy (suction on day 9). At the time of suction abortion, three of the four women who withdrew had reached gestational ages greater than 6 weeks without fetal cardiac activity. Two women (3%) who received methotrexate missed all follow-up visits. The first moved out of state and reported by telephone having had a D&C in an emergency room at the onset of vaginal bleeding. The second could not be reached after the first visit.
Six of the 61 women (10%) who received methotrexate had subsequent fetal cardiac activity. In four of the six, cardiac activity was noted on day 14. The remaining two failures had negative cardiac activity on day 14 but positive cardiac activity 28 days after methotrexate. Fetal cardiac activity was described as irregular by the ultrasonographer. Both women were at ages less than 5 weeks' gestation at enrollment. The presence of cardiac activity after methotrexate differed by gestational age. In gestations less than 6 weeks, cardiac activity was observed in three of 55 women (5%); in the group from 6 to 7 weeks, the rate was three of six (50%) (P < .01).
Forty-six of the 49 women without fetal cardiac activity at visit 2 after methotrexate returned for all follow-up visits or reported completion of vaginal bleeding via telephone interview. Three of the 49 with negative cardiac activity at visit 2 missed further follow-ups. Their final outcomes are unknown, but could be assumed successful because at their final visits, all reached gestational ages greater than 7 weeks without fetal cardiac activity.
Thirty-five of the 49 women returned bleeding diaries (71%); the remaining 14 gave information by telephone interview. The mean first day of bleeding was day 19 (SD 7.8, range 5–35). The mean total days of bleeding was 9 (SD 4.0, range 4–21). One woman needed suction curettage because of incomplete tissue passage and bleeding (hematocrit from 39% to 30%). At curettage, a complete decidual cast was extracted easily from the cervical os. This woman was highly satisfied with her experience and said she would recommend medical over surgical abortion. One woman did not return until 6 weeks after methotrexate was given. Ultrasound documented an empty uterus and a sensitive pregnancy test was negative; however, the woman denied any vaginal bleeding.
The 33 women who aborted and gave information about pain medication used a mean of 5.4 (SD 5.4) ibuprofen tablets and 2.8 (SD 4.3) acetaminophen with codeine tablets during the study. Thirteen of the 33 used both medications, five of 33 used acetaminophen with codeine only, ten of 33 used ibuprofen only, and five of 33 used neither.
Although used for the treatment of trophoblastic neoplasia and known to increase the risk of first-trimester loss, 200 mg of oral 6-mercaptopurine did not induce abortion. All women who received 6-mercaptopurine returned with positive fetal cardiac activity and increased β-hCG. The standard leukemia induction dose for 6-mercaptopurine is 80–100 mg/m2 daily for 4 weeks. Although nearly double the daily induction dose, 200 mg of 6-mercaptopurine was ineffective for medical abortion. It seems unlikely that the antimetabolite effect of 6-mercaptopurine was limited by pregnancy's effects on oral absorption and volume of distribution. Surprisingly, cell division in trophoblastic tissues may be resistant to 6-mercaptopurine.
Methotrexate is an effective single-agent abortifacient at early gestational ages. Forty-eight of 61 women receiving it (79%) had successful medical abortions without surgical intervention. The success of abortion appeared to be related to gestational age. Administration was followed by the presence of fetal cardiac activity in 5% of gestations within 6 weeks of LMP, but in half of gestations between 6 and 7 weeks (P < .01). Among women earlier than 6 weeks after the LMP, 83% aborted without surgical intervention, whereas among women between 6 and 7 weeks, half aborted successfully (P = .10). Small numbers of women limited the power of the study to detect a significant difference in abortion success by gestational age.
Two women at gestational ages less than 5 weeks at enrollment had positive fetal cardiac activity on day 28, without evidence of cardiac activity on day 14. Given the very early gestational age at enrollment, day 14 may have predated the expected development of cardiac activity. Failures at such early gestational ages suggest a critical period for methotrexate effectiveness. All failures highlight the need for aggressive follow-up and full understanding of the teratogenic nature of methotrexate.
Bleeding and expulsion of fetal tissue typically began 19 days after methotrexate was given. Total days of bleeding was similar to and possibly decreased compared with methotrexate used with PGs, reported to range from 9.0 to 29 days.20,21 Women required minimal pain medication during methotrexate abortions. Reported medication use reflects the entire study interval and overestimates use during abortion. Use of pain medications was similar in the 6-mercaptopurine group who did not abort and the methotrexate group who did abort, but small numbers of subjects limited the power to detect a significant difference. The nonmasked route of drug administration may have affected the perception of pain and pain medication use. Future studies should address baseline use of pain medication during early, unwanted pregnancy and PG effects on pain during medical abortion.
Our protocol initially included baseline and repeat determinations of complete blood counts, aspartate serum transaminase, and creatinine. Among the small numbers of women tested, none was excluded from enrollment and no case of toxicity was identified. Published studies using methotrexate vary in the use of laboratory tests. Neutropenic fever has occurred after methotrexate, but was not preceded by a low WBC count.22 Given the cost, labor, and inconvenience of these tests, their value should be clearly established before routine use during medical abortion.
Methotrexate has advantages as a single agent for early abortion. Eliminating PG administration by providers in the clinic or by women at home simplifies the regimen. Methotrexate alone may be preferable in women with medical conditions sensitive to PGs, such as poorly controlled hypertension, heart disease, inflammatory bowel disease, vasculopathy, or reactive airway disease. A regimen of methotrexate alone eliminates the independent exogenous PG-mediated side effects. For protocols including PGs, high success rates with methotrexate alone are reassuring because women who fail follow-up or incorrectly self-administer PGs will probably still abort at early gestational ages.
Methotrexate is known to be cytotoxic to trophoblast, but its mechanism of action in medical abortion remains obscure. In this study, placental tissues exposed to the clinically effective methotrexate showed diminished Ki-67 activity compared with placental tissues exposed to the clinically ineffective 6-mercaptopurine, a difference not demonstrated with routine hematoxylin and eosin staining. These findings were limited by the few specimens tested and the lack of control tissues. If prospective studies identify an association between Ki-67 activity and clinical abortion success, then immunohistochemical analysis of Ki-67 activity could be used to screen potential abortifacients or to adjust dosing.
Some women prefer medical abortion. Reasons cited include avoidance of invasive surgery, the feeling that a medical abortion is more natural, increased privacy, a previous negative experience with surgical abortion, and the option of initiating medical abortion at very early stages of pregnancy.23 Current options for medical abortion include methotrexate and mifepristone augmented with PGs, as well as methotrexate alone. Specific protocols using these agents vary, and the regimen associated with the least side effects, highest efficacy, and greatest ease of clinical implementation has yet to be found. Future research may identify other agents that make safe, effective medical abortion an option for more women.
1. Riddle JM. Contraception and abortion from the ancient world to the Renaissance. Cambridge, Massachusetts: Harvard University Press, 1994:16–24.
2. Etziony MD. The physician's creed; an anthology of medical prayers, oaths, and codes of ethics written and recited by medical practitioners through the ages. Springfield, Illinois: Thomas, 1973.
3. Thiersch JB, Philips FS. Effect of 4-amino-pteroylglutamic acid (aminopterin) on early pregnancy. Proc Soc Exp Biol Med 1950;74:204–8.
4. Thiersch JB. Therapeutic abortions with a folic acid antagonist, 4-aminopteroylglutamic acid (4-amino PGA) administered by the oral route. Am J Obstet Gynecol 1952;63:1298–304.
5. Tanaka T, Hayashi H, Kutsuzawa T, Fujimoto S, Ichinoe K. Treatment of interstitial ectopic pregnancy with methotrexate: Report of a successful case. Fertil Steril 1982;37:851–2.
6. Stovall TG, Ling FW. Single-dose methotrexate: An expanded clinical trial. Am J Obstet Gynecol 1993;168:1759–65.
7. Creinin MD. Methotrexate for abortion at <42 days gestation. Contraception 1993;48:519–25.
8. Hausknecht RU. Methotrexate and misoprostol to terminate early pregnancy. N Engl J Med 1995;333:537–40.
9. Ulman A, Silvestre L, Chemama L, Rezvani Y, Renault M, Aguillaume CJ, et al. Medical termination of early pregnancy with mifepristone (RU 486) followed by a prostaglandin analogue. Acta Obstet Gynecol Scand 1992;71:278–83.
10. World Health Organization (WHO), Task Force on Postovulatory Methods for Fertility Regulation. Pregnancy termination with mifepristone and gemeprost: A multicenter comparison between repeated doses and a single dose of mifepristone. Fertil Steril 1991;56:32–40.
11. American Hospital Formulary Service drug information 1997. Bethesda, Maryland: American Society of Health System Pharmacists, Inc., 1997.
12. Bagshawe KD. Trophoblastic tumours: Chemotherapy and developments. BMJ 1963;2:1303–7.
13. Hung-Chao S, Pao-Chen W, Ts'ui-Hua H. Treatment of choriocarcinoma and chorioadenoma destruyens with 6-mercaptopurine and surgery. Chin Med J 1963;82:24–38.
14. Schammel DP, Blockage T. p53, PCNA, and Ki-67 in hydropic molar and nonmolar placentas: An immunohistochemical study. Int J Gynecol Pathol 1996;15:158–66.
15. Du Bois D, Du Bois EF. A formula to estimate the approximate surface area if the height and weight be known. Arch Intern Med 1916;17:863–71.
16. Cattoretti G, Becker MH, Key G, Duchrow M, Schluter C, Galle J, Monoclonal antibodies against recombinant parts of the Ki-67 antigen (MIB 1 and MIB 3) detect proliferating cells in microwave-processed and formalin-fixed paraffin sections. J Pathol 1992;168:357–63.
17. Gown AM, deWever N, Battifora H. Microwave-based antigenic unmasking: A revolutionary new technique for routine immuno-histochemistry. Appl Immunohistochem 1993;1:256–66.
18. Hsu SM, Raine L, Fanger H. Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques—a comparison between ABC and unlabelled antibody (PAP) procedures. J Histochem Cytochem 1981;29:577–8.
19. Lehr HA, Hansen KD, Coltrera MD, Russ JC, Gown AM. Photoshop-based image analysis for the semiautomated assessment of Ki-67-defined proliferative activity in the routine diagnosis of breast cancer. Appl Immunohistochem 1996;4:117–27.
20. Schaff EA, Penmetsa U, Eisinger SH, Franks P. Methotrexate. A single agent for early abortion. J Reprod Med 1997;42:56–60.
21. Creinin MD, Darney PD. Methotrexate and misoprostol for early abortion. Contraception 1993;48:339–48.
22. Isaccs JD, McGehee RP, Cowan BD. Life-threatening neutropenia following methotrexate treatment of ectopic pregnancy: A report of two cases. Obstet Gynecol 1996;88:694–6.
23. Winikoff B. Acceptability of medical abortion in early pregnancy. Fam Plann Perspect 1995;27:142–8.